You have (3) parallel PUs on each I2C signal? If you really need stiff 1.6k I2C PUs, you could just use a single 1.6k R on each signal and delete the (4) redundant ones. That might afford you a cleaner placement where you could avoid your x-over?
There are many more oddities on this layout. Caps with odd footprints sprinkled all over the board and funny track widths plus jerky routing.
No.
If you route connection at second layer you will have one net.
If you then don’t have this second layer and replace the connections made there by 0Rs then you have what you wanted.
Long time ago (before I start using KiCad in 2017) I got that idea and used third layer for 3 connections intended to be made by wires. Thanks to this DRC not complains. After ordering 2 layer board I printed third layer as a instruction where to make the wire connections.
I have added single pads instead of vias to have them not mask covered. I have done it at PCB, but it was not KiCad. In KiCad I would probably add them at schematic.
Thanks… I am pretty new to KiCad and schematic design; can you please explain further?
What do you suggest should be done to improve the quality of the PCB and also the track width what is the recommended widths for different signals?
That kind of question is not KiCad specific and really beyond the intended scope of this Forum . . .
Alright… Understood 
1 Like
I assumed this was just a test project for learning to work with KiCad. The first 5 or so projects I did were not intended to be real (working) projects, but only to learn how the design filosofy in KiCad works and where the buttons are. Concentrating on that, without having to divide your attention over getting a real project working is a good start. Once you’re fairly familiar with KiCad, you can change more of your focus to the circuit and refine PCB design.
1 Like
One thing to consider is your grounding. I think the one layer board will be a problem for grounding.
Ideally, you should make the back layer a solid copper plane connected to GND.
If you really can’t have a back layer (and that’s hard to imagine in 2025) the next best thing is probably to fill all the spaces on the top layer with GND plane. But then you’ll have to connect all the interrupted parts of that GND plane using more zero ohm resistors, or wire bridges on the bottom.
Without proper grounding you have a much bigger risk of signal integrity problems and probably a lot more EMI, which may not be great for the performance of your LORA module.
The many experts here may have more professional viewpoints on this topic, but I’d say grounding is a problem with your current design.
Regarding your question of track widths, I’d just make everything 0.3 or 0.25mm, which is enough for signals and low current power lines, and incurs no extra fabrication costs.
If you’re expecting higher currents, you can route power traces as 0.5mm and signal traces as 0.25mm.
If your currents are expected to be more than 2A you’ll have to start to think about power design and thermal considerations.
1 Like
Yes, absence of a GND plane is a quite big issue, but it can be done. Cheap FR2 boards are made en masse, and these product appertly conform to EMC regulations.
Etching at home is still a thing. It’s not a cost saving, and quality does not compete with dual layer PCB’s with drilled and plated holes and soldermask and silkscreen on both sides, but the speed still remains. You can go from a finished PCB design to starting to solder in an hour or less. And that’s also why there are commercial products (such as LPKF) for milling PCB’s. The quick turnaround is important for prototypes during the design phase.
When you make your own PCB’s, you can also use the good old star grounding. I also do this with prototypes on matrix PCB’s (I do draw a schematic, and use the PCB editor for footprint placement, so I know I reserve enough space for stuff.) I drill a big hole in the PCB, and that is where all the GND wires go to.
1 Like
(I used Google translate).
Perhaps the translation is bad, but zero ohm SMT resistors are very cheap, similar to other resistor values in the same package. I would think that a fuse is significantly more expensive.
I used wire links, placed by machine in the old days. These looked a bit like a THT resistor, but made as wire with a short sleeve and supplied as a bandolier
1 Like
Das ist die Frage, ob das Problem mit dem 0-Ohm-Widerstand auf dem PCB einfacher gelöst werden kann oder ob der Preis für den 0-Ohm-Widerstand enstcheidend ist. Man kann ja mal die Preise vergleichen und dann entscheiden.
Nie tylko cena jest problemem. Bezpiecznik, aby mógł zadziałać potrzebuje aby się w nim wydzieliła wystarczająca ilość energii i dlatego nie może mieć rezystancji równej 0, a rezystor 0R może mieć rezystancję tak niską na ile technologia pozwoli. Jeśli twoim celem jest połączenie dwu punktów to jaki jest powód aby wstawiać tam jakąś zbędną rezystancję dającą dodatkowe spadki napięcia. Zdarzyło mi się użyć rezystorów 0R kilka razy na liniach VCC 3V3 przy prądach rzędu 200mA. Jaki będzie spadek napięcia na bezpieczniku użytym w tym miejscu?
1 Like
We have had long and serious discussion about the language policy in this forum, and the result was that English is the official and main language here. We can allow non-English every now and then, but if someone wants to continue in other than English, the minimum requirement is that an English translation (made by some automatic service is OK) is attached to each post.
I recently made some single layer prototype boards and also needed some jumpers. The way I got round it was in KiCad to design a double layer board but the tracks in the back layer I later manually soldered wires instead of etching
3 Likes
This is the standard way to solve the problem.
Now that we have all had plenty of time to go off topic (KiCad being the correct topic) I think we can draw this thread to an end.
This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.